Swan Chain Network Status Report — March 3, 2026

[flyworker]

Swan Chain Network Status Report — March 3, 2026

Data captured from provider.swanchain.io on March 3, 2026.

---

Network Overview

Metric	Value
Total Computing Providers	64
Listed Active FCP	50 (only 6 genuinely operational)
Listed Active ECP	14 (only 3 genuinely operational)
Total GPUs Registered	166
Provider Locations	19 listed (3 with operational providers)
Total Computing Units (CU)	1,255.5
Total ZK Tasks Completed	1,637,229
Total Deployments	706,781
Total SWAN UBI Paid	22,694,060.70 SWAN
Total SWAN Collateral Locked	6,018,750.75 SWAN

---

Genuinely Operational Providers

Out of 44 listed FCPs and 3 ECPs, only 6 unique providers are genuinely working (uptime > 0%). Three of them run dual FCP+ECP roles.

Active FCPs (Fog Computing Providers)

#	Name	Address	Uptime	GPUs	Region	CU	Deployments
1	nocall	0x4dF...3B076	99.42%	TITAN RTX, GTX 1050 Ti, TESLA P4, A4000	California, US	112.7	147,039
2	(unnamed)	0x8B1...B8695	100%	TESLA P4, RTX 3070, TITAN RTX	California, US	85.4	73,639
3	(unnamed)	0x5aC...a72d5	99.42%	TESLA P4, RTX 3080	Hunan, CN	22.7	35,711
4	(unnamed)	0xE63...13a19	100%	TESLA P4, GTX 1050 Ti	California, US	41.8	54,836
5	mqm	0x614...246b1	99.42%	TESLA P4, RTX 4060, GTX 1050 Ti	California, US	42.9	50,950
6	(unnamed)	0xf95...3C58c	58.29%	RTX 4070, RTX 5090	Tokyo, JP	8.6	3,178

The remaining 38 FCPs listed as "Online" all show 0% uptime -- registered but not serving workloads.

Active ECPs (Edge Computing Providers)

#	Name	Address	Completion Rate	GPUs	Region	CU	Total Tasks
1	myh	0x5aC...a72d5	91.3%	TESLA P4, RTX 3080	Hunan, CN	22.7	20,986
2	node2	0x8B1...B8695	90.84%	TESLA P4, RTX 3070, TITAN RTX	California, US	85.4	19,325
3	node3	0xE63...13a19	85.09%	TESLA P4, GTX 1050 Ti	California, US	41.8	12,720

Addresses 0x5aC, 0x8B1, and 0xE63 run both FCP and ECP roles on the same hardware.

---

GPU Types Actually In Service

GPU Type	In Active Providers
TESLA P4	5 providers
NVIDIA TITAN RTX	2 providers
NVIDIA GTX 1050 Ti	3 providers
NVIDIA RTX 3080	1 provider
NVIDIA RTX 3070	1 provider
NVIDIA RTX 4060	1 provider
NVIDIA RTX 4070	1 provider
NVIDIA RTX 5090	1 provider
NVIDIA A4000	1 provider

---

Regions With Operational Providers

Region	Active Providers	Role
California, US	4 (nocall, 0x8B1, 0xE63, mqm)	FCP + ECP
Hunan, CN	1 (myh / 0x5aC)	FCP + ECP
Tokyo, JP	1 (0xf95)	FCP

---

All Registered GPU Inventory (Rentable)

Total Available GPUs: 68 / 291

GPU Model	Available / Total	vRAM	Price ($/hr)
H100	40 / 64	80Gi	$1.20 - $1.40
A100	10 / 12	40Gi	$1.20
RTX 4090	32 / 43	24Gi	$0.39
A6000	55 / 56	48Gi	$0.31 - $0.41
RTX 3090	10 / 14	24Gi	$0.23 - $0.28
RTX 3070	18 / 20	8Gi	$0.14 - $0.19

Note: The H100, A100, RTX 4090, A6000, and RTX 3090 rentable inventory is associated with providers showing 0% uptime. These GPUs are listed but may not be genuinely available.

All GPU Types by Unit Count

GPU Type	Count
TESLA P4	37
RTX 4070 Ti	36
GTX 1050 Ti	24
RTX 3080	17
RTX 4090	11
L40S	8
TITAN RTX	8
A4500	8
RTX 3070	3
A4000	2
RTX 3090	2
RTX 3080 Ti	2
RTX 4060	2
RTX 2080 Ti	2
A100-PCIE-40GB	1
RTX 4070	1
RTX 3060	1
RTX 5090	1

---

Resource Utilization

Resource	Used	Free	Total	Utilization
GPU	61	105	166	36.7%
CPU (cores)	331	3,023	3,354	9.9%
Memory	3.11 TB	17.37 TB	20.48 TB	15.2%
Storage	20.14 TB	53.58 TB	73.72 TB	27.3%

---

All Provider Regions (by CPU cores)

Region	CPU Cores
Shanghai, CN	1,120
Saxony, DE	768
California, US	384
Quebec, CA	208
Kwai Tsing, HK	128
North Carolina, US	128
Sichuan, CN	96
South West, SG	96
Tokyo, JP	68
North West, SG	64
Virginia, US	64
Bavaria, DE	64
Hunan, CN	54
Grand Est, FR	40
Kowloon, HK	32
Central Singapore, SG	20
New York, US	20

---

UBI Cost Analysis

SWAN Token Price: $0.0005016 (March 3, 2026)
UBI Distribution Contract: 0xE7b1Dc0F5C33214c5dc213ea6215d5b10508aBc8

Metric	SWAN	USD
Daily UBI Payout	58,369	$29.28
Monthly UBI	1,751,070	$878
Yearly UBI	21,304,685	$10,686
Total UBI Paid to Date	22,694,061	$11,383

Where Daily UBI Goes

Recipient	SWAN/day	USD/day	Share
6 active providers (working)	21,572	$10.82	37%
38+ idle providers (0% uptime)	36,797	$18.46	63%
TOTAL	58,369	$29.28	100%

63% of daily UBI goes to providers with 0% uptime.

Per Active Provider UBI Earnings

Provider	CU	Uptime	SWAN/day	USD/day
nocall	112.7	99.42%	5,239	$2.63
0x8B1 (node2)	85.4	100%	3,970	$1.99
mqm	42.9	99.42%	1,994	$1.00
0xE63 (node3)	41.8	100%	1,943	$0.97
0x5aC (myh)	22.7	99.42%	1,055	$0.53
0xf95 (Tokyo)	8.6	58.29%	400	$0.20

---

Key Findings

1. Only 6 unique providers are genuinely operational out of 64 listed. That is 9.4% of listed providers actually serving workloads.

2. Dual-role providers are the backbone: 3 providers (0x5aC, 0x8B1, 0xE63) run both FCP and ECP, handling the majority of ZK tasks and deployments.

3. High-value GPU inventory is idle: The H100 (64 units), A100 (12 units), RTX 4090 (43 units), and A6000 (56 units) are all listed under providers with 0% uptime. These GPUs appear available on the resource page but the providers hosting them are not demonstrably operational.

4. Geographic concentration is extreme: All active providers are in just 3 locations -- California US (4), Hunan CN (1), and Tokyo JP (1). The 14 other listed regions have zero operational providers.

5. TESLA P4 dominates actual workloads: The most common GPU in genuinely active providers is the TESLA P4, a datacenter card from 2016. The flagship consumer/prosumer GPUs (H100, A100, 4090) are not in active use.

6. 63% of UBI is wasted: At $29.28/day total UBI, $18.46/day goes to idle providers with 0% uptime. This amounts to ~$6,737/year in SWAN tokens distributed to providers doing zero work.

7. One RTX 5090 spotted: A provider in Tokyo, JP has an RTX 5090 listed -- notable as a next-gen GPU, though this provider has only 58.29% uptime.

---

Recommendations

1. Redefine "Active" status: The dashboard showing 50 active FCPs when only 6 work is misleading. Recommend adding an "Operational" badge that requires uptime > 50% in the trailing 7 days.

2. Prioritize quality over quantity: With Swan 2.0 Inference Cloud launching, having 6 reliable providers is more valuable than 64 mostly-phantom ones. The SIP-002 contribution-based rewards model will naturally address this.

3. Regional expansion is critical: 3 active regions for a global inference network is insufficient. Consider targeted incentive programs for providers in Europe, Southeast Asia, and South America.

4. GPU modernization: The active fleet is dominated by TESLA P4s. For competitive AI inference (LLM, image generation), the network needs more H100/A100/4090 capacity from providers that are actually online.

5. UBI reform: Redirect the 63% of UBI currently going to idle providers toward contribution-based rewards for providers that actually serve inference workloads (per SIP-003).

---

Report generated from live dashboard data on March 3, 2026.